Research On Abnormal Pattern Mining Algorithm In Complex Network

Intelligence analysis is a cross-cutting field that has developed over the years.Based on complex network-related algorithms,it conducts intelligence analysis and research on the perception,understanding,and prediction of people(accounts),things,organizations,true and false events,and their relationships.From large-scale social relationship networks based on human electronic footprint in social/physical/cybersp-ace The key algorithm of the understanding module in the intelligence analysis framework IAF is to quickly identify and detect events through frequent or abnormal pattern mining of human interaction.This paper focuses on the mining of abnormal patterns,and carries out the following work:First,a general frequent pattern mining algorithm GCFPM is proposed.The main idea of the algorithm is to traverse the whole graph based on depth-first search to obtain a linear sequence of graph coding composed of five-tuple coding.The idea of right-most path expansion is used for expansion and matching,and frequent patterns are quickly mined.After optimization,this method can effectively improve time efficiency and save space cost.Secondly,an abnormal pattern mining algorithm GCAPM for complex networks is proposed.The algorithm is based on a frequent pattern mining algorithm that improves efficiency in the first step,and provides a formal definition of the corresponding"abnormal pattern",including defining two support thresholds supp_?1and supp_?2to mine abnormal patterns and visualize the results.Compared with frequent patterns,analyze the possible causes of abnormal patterns in different situations.Finally,an empirical analysis is made on the above research methods.The transaction data of a commercial bank is used to construct a financial transaction network,and anomalous patterns are discovered from it.Aiming at the abnormal behaviors that may occur during bank transactions,it is verified that the algorithm can mine such abnormal patterns,which is helpful for discovering potential intelligence in the financial field and for further intelligence analysis by researchers.In summary,based on the complex network theory and the abnormal event detection function of the understanding module in the intelligence analysis framework IAF,the above two pattern mining algorithms are completed.On the basis of improving the efficiency of the algorithm,an empirical analysis using real data of a commercial bank proves the application value of the abnormal pattern mining algorithm,and gives an explanation of the possible abnormal behavior in practical applications,which has important practical significance.